Home Industries Platform Nx·Studio Flagship Nx IDE Cognitive Engine Trust Center Pricing Capabilities About Us Leadership Team Careers Contact Us
Enterprise automation. Owned by you.

Scope one process, assign its owners and run workers your teams control.

Enterprise automation. Owned by you.

Make AI an operational asset.Build a workforce you own.

Assign each routine process an owner and approvers, then let a digital worker carry the volume.Your process owner scopes each worker in Nx·Studio alongside engineering. Approvers walk through their checkpoints in trial runs, then IT deploys the code where your teams already work.

Plan the rollout in stages. Agree volumes and exception targets, name an approver for each checkpoint, and run trials on generated test cases. Go live only when the team accepts the results.

The rollout, in four steps

  1. 01

    Define

    Set scope and owner

  2. 02

    Configure

    Add rules and approvers

  3. 03

    Trial

    Check it on test data

  4. 04

    Operate

    Deploy on your systems

Define. Configure. Trial. Operate.
111

Portals delivered so far

Complete business applications your teams can rebrand and operate.

333

Digital workers built

Each one handles a single defined process, with the rules and approvers its owner sets.

17

Sectors covered

Covering finance, healthcare, manufacturing, logistics, energy, legal and automotive.

Each count is calculated from the portal and worker records listed on this site. A digital worker is an operational asset in code: it runs one defined process under your rules and approval checkpoints, and your organization holds the full source.

Products by team

Give each team the product it will work in.

Owners, engineers and case reviewers. Every worker arrives as code you hold.

01 / 04Work your teams hand off

Digital Workers

Hand repeatable work to software your team manages day to day. Each worker has a named owner, fixed rules and approval checkpoints, and exceptions return to the person responsible.

  • One owner per worker
  • Trial, then live use
  • Code your team holds
  • Hosted where IT decides

In this example

Illustrative rollout: three teams each hand one task to a worker, from claims cover checks to supplier files and work orders. Every exception returns to its owner.

  • Claims

    Cover checked per claim

  • Onboarding

    Supplier files checked

  • Maintenance

    Work orders booked

Your team, in charge

Exceptions go to their owner

How teams run digital workers
02 / 04Where owners design the flow

Nx·Studio

Gives process owners and engineers one shared workflow. Compare scored designs and estimated run costs, trial each approval checkpoint with its approver, then take delivery.

  • Designs compared, scored
  • Approvers in the trial
  • Version history kept
  • Delivered as a ZIP project

In this example

Illustrative design: a manager’s one-line request becomes a three-step approval flow in Nx·Studio, tested on generated cases, with every version kept.

Route purchase requests to the right approver
  1. New request

    From the portal

  2. Budget checked

    Against cost center

  3. Routed

    To the named approver

Saved as version 4History kept for review
Nx·Studio for owners
03 / 04For the engineering team

Nx IDE

Gives your engineering team AI specialist roles in VS Code for delivery and modernization work. One role plans, another codes, others test and review, and by default nothing changes in the editor until an engineer approves.

  • 31 modes in all
  • Rules: allow, ask, deny
  • Models the team approves
  • Checkpoints to undo work

In this example

Illustrative change: the team’s Plan, Tester and Reviewer roles add a supplier lookup in turn, and by default the editor holds it for an engineer’s approval.

supplier-lookup.planHeld for approval
# Plan: add a supplier lookup
task:
why: requested by procurement
model: as configured
work:
- add the lookup form
- test with sample IDs
- request a review
release: when approved

The team’s AI roles

Plan

Sets out the work

Tester

Tests, never edits

Reviewer

Second look at code

Engineer approvesHolds by default
Nx IDE for the team
04 / 04For routine case decisions

Cognitive Engine

Learns decision rules from outcomes your reviewers confirm. No rule decides until it passes a held-out exam; unfamiliar cases stay with your reviewers. Published results are synthetic.

  • Passes an exam first
  • Decides without AI calls
  • Reviewers get new cases
  • Pulled when proven wrong

In this example

Illustrative decision: a qualified rule settles a routine motor reading with no AI call. Unfamiliar readings go to the team’s case owner, and AI helps only if enabled.

Motor current: within its normal band?Verdict
  • Qualified pattern

    No AI call needed

  • AI, if enabled

    Opt-in only

  • Case owner

    Takes new case types

ObserveQualifyOperate
How reviewers work with it
Where your team's time goes

Start with one team's workload.

Record today's volumes, handling time and exception rates for one team. Name the process owner and approvers, then compare the same figures in test runs before any wider rollout.

01 / WORKLOAD

Move routine volume off the team

Workers take the steps your rules settle with no model call; each run logs its own AI cost. Report queue size, handling time and cost per case to the team every week.

What each run costs
02 / QUALITY

Keep people on the judgment calls

Your reviewers receive the cases that fall outside the rules, so their time goes to decisions rather than re-keying. Track the exception rate for one process, then add a second.

Read the four steps
03 / OVERSIGHT

Test each approval pause first

Team leads watch each workflow run on generated test cases, with outside systems simulated and approval checkpoints active. Read each run's decision record before anything connects to live systems.

The four safeguards
04 / CONTINUITY

Runs on after the handover

Your IT team takes delivery of the full source code and runs it on systems you choose, with the model vendor you contract directly. Day-to-day running needs no license or meter; Managed Delivery is optional.

Owning or renting
Illustrative example

One payables team over a typical month.

Take a payables team that matches 2,000 invoices a month to purchase orders by hand, six minutes each: 200 hours of routine work.

TEAM BASELINE
2,000invoices a month
6 mineach, keyed in
200 hof team time monthly
01 / SETTLED
1,500

Invoices the worker could clear for the team

If 75% meet every matching rule the team set.

Settled by fixed rules75%
02 / TO REVIEW
500

Invoices still worked by your reviewers

A quarter fall outside the rules and need a decision.

Back to a reviewer25%
03 / REASSIGNABLE
150

Hours the team can reassign

Gross capacity: setup, review time and hosting are not deducted.

Of today's 200 hours75% at best

A sample calculation, not a customer result. It assumes 75% of invoices clear the team's rules and that each one cleared saves six minutes of a clerk's time. Setup, review and run costs are excluded; test each input against your own team's figures.

Who does what

Clear owners at every step of the rollout.

Your process owner leads each stage with engineering, approvers and IT, and accepts it before the next one begins.

  1. Stage 1 / Define

    The process owner sets the scope.

    The owner agrees boundaries, inputs, rules and expected results with the team. Procedures and policies can be uploaded, and Nx·Studio proposes designs based on how the work actually runs.

  2. Stage 2 / Configure

    Set who approves what.

    Owners and engineers compare the proposed designs together. On the visual canvas, the team adjusts steps and fixed rules and assigns an approver to every checkpoint; edits can be undone.

  3. Stage 3 / Trial

    The team trials it on test cases.

    Nx·Studio generates test cases and simulates the systems involved. Owners review outputs and exceptions, approvers see each stop that waits for them, and engineers then test each real connection.

  4. Stage 4 / Operate

    IT deploys it; the owner runs it.

    IT takes delivery of the complete Python project as a ZIP, sets up credentials and hosting, and tests it against your systems. Day to day, the process owner reviews its operating records; Nx·Studio is not needed.

A forecast worth planning for
Industry outlook / 2027

Gartner expects organizations to cancel more than 40% of their agentic AI projects by the close of 2027. The causes it cites are mounting costs, business value that remains in doubt, or risk controls that prove inadequate.

Cover all three in the rollout plan.

01
Escalating costs

Review each design's run-cost estimate, then the AI spend each trial records.

02
Unclear business value

Agree the process owner and the result measure before the first trial.

03
Inadequate risk controls

Name the approvers and test the rule checks before live operation.

After export from Nx·Studio, your own team runs the code.

Who runs it day to day

Vendor-run or team-run? Settle it at the start.

Compare who hosts it, what it costs to run, who can inspect it and what happens at contract end.

Rental model
Location

Runs where the vendor decides.

Running cost

Set by the vendor's prices and usage terms.

Code access

Confined to the vendor's own reports.

Contract end

The service can stop with the contract.

Ownership model
Location

Runs on systems your team manages.

Running cost

Hosting, model fees and upkeep in your budget.

Code access

Your team reviews the workflow and code.

Contract end

Keeps running; your team maintains it.

Owned code puts every change through your change process.

Plan who covers hosting, integrations, model accounts and upkeep.

How fees are split

Your team receives the whole worker. Signed, then checked in-house.

Each package exported from Nx·Studio contains a signed manifest listing its models, agents and dependencies. It is created by default, and your security reviewers check it before deployment.

Which product each team runs

Assign the work. Set the controls. Track the results.

Digital workers handle the routine steps of a process, and your team designs and trials them in Nx·Studio. Nx IDE is where your engineers deliver changes, in VS Code. Cognitive Engine takes repeat decisions once its rules qualify, and returns the rest to your reviewers.

01DefineFlagship

Nx·Studio

Your process owner leads the work. The team compares up to five costed designs, sets who approves each checkpoint and reviews trial runs on test data; then engineers take delivery of the code.

  • Up to five designs to compare
  • Approvers set per checkpoint
  • Trial results first
Nx·Studio at work
02Deliver

Nx IDE

Your engineering lead keeps change control. Each request is planned, coded and independently reviewed in VS Code, with the model vendor the team approved and action limits the lead sets.

  • Plans approved before coding
  • Testing and review apart
  • Impact rated up front
Nx IDE at work
03Qualify

Cognitive Engine

Your decision owner stays accountable. Reviewers confirm outcomes, the engine learns rules from them, and a rule may act only once it scores perfectly on held-back cases. Anything unfamiliar comes back to the queue, or to AI if you allow it.

  • Proven rules, no model call
  • Opt-in AI advice, also examined
  • Behind a separate API
The engine at work
04Audit & Retain

The Foundation

Managers, risk and procurement approve from the same evidence: workflow versions, trial outcomes and the checks run ahead of export. They then check the delivered code and its signed component list.

  • Every export passes a build check
  • Rules and approvers on record
  • A signed AI manifest
The Foundation at work
Systems in scope

List the systems each worker touches,
then plan who connects them.

Integration starts from a catalog listing thousands of APIs. Starter code covers SAP, Workday and similar systems; schedule engineering time to finish each connector, hold its credentials and test it in the workflow.

Microsoft Teams
Slack
SAP
Salesforce
Oracle EBS
Workday
ServiceNow
Email
Nxgyle

Logos are illustrative; no customer or partner relationship is implied. Engineering confirms each connection before live use.

The Foundation

What your team checks before use.

Process owners, approvers and IT each review a different part before real cases arrive. The four mechanisms give each of them something concrete to check.

  1. 01
    Rehearsal before reality

    See how cases flow first.

    Each worker processes generated cases shaped like your work, with outside systems simulated. Your team checks outputs and approval pauses before anything live is connected.

  2. 02
    Rules you can rely on

    Your rules, run as checks.

    Thresholds and policies become fixed checks wherever they can be written that way. A process owner accepts each drafted rule; from then on it gives the same result without an AI call.

  3. 03
    Blueprints you can read

    A plan the whole team can read.

    Steps, routing and approval checkpoints appear in one readable blueprint. Versions are kept, so a manager can compare two before approving a change, or branch a new one.

  4. 04
    Proof in every export

    IT verifies every component.

    Each export includes an AI bill of materials, a signed manifest naming each model, agent and dependency. Your IT team verifies it offline before the worker goes into service.

Who controls each step

Four control points your team sets up.

Your process owner, risk lead and IT owner sign off these points ahead of live operation.

See the control checklist
Configured up front
  • 01 / 04
    TRIAL

    Test before release

    Your process owner reviews trial results on generated test cases, with outside systems simulated. IT still tests the live connections separately.

    Outside systems simulated
  • 02 / 04
    POLICY

    Rule checks in the flow

    Policy owners confirm each rule Nx·Studio turns into a fixed check, with no AI call. Gaps it flags go back to them; AI steps get separate tests.

    Fixed rules, no model call
  • 03 / 04
    AUTHORITY

    Approval checkpoints

    Name an approver for each step that needs a person. The worker pauses at that checkpoint until the approver decides.

    Approvers you name
  • 04 / 04
    HISTORY

    Record each version

    Each workflow version stays on file and every approval is recorded, giving team leads a trail for change reviews.

    Change history kept
Who does what

Divide the duties.
Agree each step
before it goes live.

Decide approvers, permitted data and upkeep duties before the worker takes real cases.

Trust Center details
  1. 01
    UPKEEP

    Agree who runs and patches it

    The source code is delivered to you and runs on systems you choose, with no Nx·Studio dependency. Assign credentials, hosting, patching and maintenance to named people in IT.

  2. 02
    TEST INPUTS

    Set the test data rules

    Test runs use generated cases, so evaluating a workflow does not require live records. List what the tests cover and which systems are simulated, and check the model vendor's data terms before sensitive data is used.

  3. 03
    APPROVAL DUTIES

    Name an approver per pause

    Each approval checkpoint gets a named approver, and the worker waits for their answer. A reviewer inbox comes in the export for your developers to connect; Slack delivery is optional.

  4. 04
    AUDIT HISTORY

    Keep the history reviewable

    Workflow versions, the rules taken from them and each run's decision record stay available, and every approval is logged. The signed manifest confirms the package is unaltered; it does not certify compliance.

By sector

Choose a pilot process from your sector's work.

Each sector lists processes a team could start with. A readable workflow, test runs on generated cases and delivered source code let process owners, risk and IT staff agree on a pilot before it goes live.

Financial Services

Risk review, onboarding and payment checks for operations.

Healthcare

Patient intake, claims and rostering for admin teams.

Supply Chain

Shipment delays, stock counts and reorders for planners.

Manufacturing

Maintenance, quality holds and shift plans for plant teams.

Illustrative
  • Loan document review
  • Client onboarding
  • Payment exceptions
See the sector
Illustrative
  • Patient intake forms
  • Claims follow-up
  • Shift rostering
See the sector
Illustrative
  • Late shipment follow-up
  • Cycle counts
  • Reorder approvals
See the sector
Illustrative
  • Maintenance requests
  • Quality holds
  • Shift handovers
See the sector

Each sector's examples show what the platform can generate. They are illustrations, not customer references.

Common questions

Questions program teams ask.

Ask your own question

Pick one high-volume process with clear rules and a named owner, such as KYC reviews or supplier-invoice reconciliation. Fixed rules settle its routine steps, AI handles judgment, and approval checkpoints sit wherever a person must decide. Your team then runs the delivered code.

Start with a priority process

Define the value. Validate the work.Own the code. Control the change.

Bring one process, its owner and its approvers.
We will plan the trial and the rollout with your team.

  • Code your team runs

    Readable source, tests and docs your engineers maintain.

  • Checked contents

    Your reviewers can confirm no exported file was altered.

  • Hosting your IT picks

    Your cloud or servers, under your IT procedures.